2016/11/07 by Minh‐Tan Pham, Pham, Minh-Tan, Grégoire Mercier +5
Computer Science · #Advanced Image and Video Retrieval Techniques #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Image Retrieval and Classification Techniques #Medical Image Segmentation Techniques
paper · pdf · doi:10.48550/arxiv.1611.02102
openalex publication_date 2016/11/07 · openalex created_date 2022/10/02 · openalex updated_date 2026/07/28
A novel efficient method for content-based image retrieval (CBIR) is\ndeveloped in this paper using both texture and color features. Our motivation\nis to represent and characterize an input image by a set of local descriptors\nextracted at characteristic points (i.e. keypoints) within the image. Then,\ndissimilarity measure between images is calculated based on the geometric\ndistance between the topological feature spaces (i.e. manifolds) formed by the\nsets of local descriptors generated from these images. In this work, we propose\nto extract and use the local extrema pixels as our feature points. Then, the\nso-called local extrema-based descriptor (LED) is generated for each keypoint\nby integrating all color, spatial as well as gradient information captured by a\nset of its nearest local extrema. Hence, each image is encoded by a LED feature\npoint cloud and riemannian distances between these point clouds enable us to\ntackle CBIR. Experiments performed on Vistex, Stex and colored Brodatz texture\ndatabases using the proposed approach provide very efficient and competitive\nresults compared to the state-of-the-art methods.\n